Active Pixel Sensors in Medical and Biologi

The application of Large Area Active Pixel Sensor (LAS) to high resolution

Nuclear Medicine imaging

Bob Ott Physics Department

Institute of Cancer Research and Royal Marsden Hospital

What is an active pixel sensor (APS)

• A silicon wafer based sensor similar to a CCD but with the potential for intelligent processing in each pixel

• APSs are based on mainstream CMOS technology and have generic benefits of low-power, high-speed, cost-effectiveness, flexibility and high levels of on-chip integration.

Other APS propertiesBack thinning allow low energy beta/gamma/X-ray

radiation to be detected

Devices can be used either as direct sensors or coupled to phosphors for example.

Devices can be read out through special FPGA based interfaces to allow data acquisition onto a PC or laptop.

Potential for radiation hardness

Potential for ROI read-out

The MI-3 research consortiumMI-3 stands for Multidimensional Integrated Intelligent

Imaging

MI-3 was a four-year £4.5m Basic Technology project funded by UK-RC to advance the capabilities and application of APSs for a raft of scientific and technological endeavours

Consortium members Universities of Sheffield, Liverpool (2 groups), Glasgow, Surrey, York, Brunel, UCL and

Rutherford Appleton Laboratories MRC Laboratory for Molecular Biology, Institute of Cancer Research.

MI-3 APS applications Medical and biological imaging (UCL, Surrey, ICR)Space science (Brunel)Bioarray imaging (Liverpool)Electron microscopy (Cambridge)Particle physics (Liverpool, Glasgow)Synchrotron radiation applications (Sheffield)Materials science (York)

Detector Design and testing mostly done at RAL/STFC

Properties of APSs used in this project

# Pixels Pixel size(µm)

System noise (e)

Frames/s

Startracker 525 × 525 25 >100 10

Opic 64 × 72 40 >100 >3700

Vanilla 520 × 520 25 34-52* 1-100**

LAS 1350 × 1350 40 40-62* 20**

*Indicate the noise range which is dependent on the reset type.

**The rates depend on whether full-frame or ROI read-out is used.

High-resolution Nuclear Medicine imaging requirements

Pixel sizes of 50-100 microns

Coupling to CsI(Tl) phosphors to detect 140keV photons

Low noise needed as signal is small

>100 cm2 sensitive area

Kcps photon counting capacity

On-chip pulse and cluster analysis for signal selection and noise reduction

High-resolution gamma camera

APS coupled to pixelated CsI(Tl) phosphor via optical fibre stud

High-resolution gamma cameraprofile across slot

0

20

40

60

80

100

0 1 2 3 4 5 6

mm

coun

ts p

er p

ixel

5mm wide slot

System resolution is ~1mm FWHM

Using 2mm thick CsI(Tl) scintillator with 400µm elements coupled to Vanilla

LASPixel size 40 µm x 40 µm

1350 x 1350 pixels

54 mm x 54mm stitched sensor

Read-out noise between 40e and 62e

73% fill factor

Region of interest read-out available

Maximum frame rate 20s-1

QE at 520 nm ~22% (Vanilla >60%)

Average QE between 500nm and 800nm ~16%Bohndiek et al, IEEE TNS, 56, 2938-2946, 2009

Dark image taken with LAS

Horizontal profile

X-ray resolution phantom imaging

0.1 lp mm-1

0.25 lp mm-1

Taken with a 99mTc flood source

0

0.2

0.4

0.6

0.8

1

1.2

0 0.1 0.2 0.3 0.4 0.5 0.6

Spatial frequency

MTF

LAS coupled to unsegmented 2mm thick CSI(Tl) flood source

FWHM of ~ 2.5mm

LAS flood image of 20mm x 20mm segmented CsI(Tl) scintillators

400µm elements 350µm elements

LAS image of crossed lines thru 3mm pinhole

LAS image of 5mm diam sphere thru 3mm pinhole

Problems and solutions• The signal from Tc-99m is small so that

segmented CsI (Tl) is important

• Higher QE (from <20% to >60%) essential

• Processing to remove fixed pattern noise improves image quality

• Larger CMOS pixels with low noise preferred

• Larger area than LAS needed – see Dynamite

Active area 120 mm x 120 mm

Two pixel sizes – 50 µm and 100 µm

15 frames/s for full read-out

Faster for ROI read-out

Pixelated CsI(Tl) scintillator available

Dynamite – coming soon

ConclusionsAPSs could make a contribution to Nuclear Medicineimaging

Noise level requirements are similar to CCDs but with faster readout speed

Fixed pattern noise dominates over thermal noise

On-chip intelligence for noise filtering and allow ROI selection

Large area devices essential for this application

Dynamite will be very interesting